Equipment failure sign notification device

By designing a device for bad sign notification, using action logs and bad sign rules to detect bad signs of the equipment, we can implement response measures before bad signs of the equipment occur, solve the problem of the problem of inadequate notification of bad signs in the existing technology, and improve the equipment Reliability and availability.

CN116569183BActive Publication Date: 2024-11-12MITSUBISHI ELECTRIC BUILDING SOLUTIONS CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202080105829.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-10-08
Publication Date
2024-11-12
Estimated Expiration
2040-10-08

AI Technical Summary

Technical Problem

The prior art cannot notify bad signs before the equipment has bad consequences, and can only analyze the causes after the equipment has bad consequences.

Method used

A device for notifying bad signs of equipment was designed, through the action log acquisition department, the bad sign rule database, the bad sign judgment department and the bad sign notification department, the action log of the equipment is detected and the content of the bad sign of the equipment is determined according to the bad sign rules, and the bad sign content is determined in Take measures before any adverse events occur.

Benefits of technology

It realizes early detection and notification of bad equipment signs, reduces the number of bad things happening, and improves the reliability and usability of the equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116569183B_ABST
    Figure CN116569183B_ABST
Patent Text Reader

Abstract

The device bad sign notification apparatus (100) for notifying a bad sign of a device comprises: an action log acquisition unit (14) for acquiring an action log of a device; a bad sign rule database (23) for storing bad sign rules for associating action log data of a device with bad sign contents of the device; a bad sign determination unit (24) for determining that there is a bad sign with corresponding bad sign contents when the action log acquired by the action log acquisition unit (14) matches the action log data in the bad sign rule; and a bad sign notification unit (25) for notifying the bad sign contents determined by the bad sign determination unit (24).
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the structure of a device bad sign notification device for notifying a device of bad signs, a method for generating an action mode for detecting bad signs of a device, a method for generating bad sign rules for detecting bad sign contents of a device, and a method for generating a bad sign inference model for inferring bad sign contents of a device. Background Art

[0002] There has been proposed a method for analyzing the cause of a failure of a device based on a log output by the device or an image of the operating status of the device (see, for example, Patent Document 1).

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2011-108154 Summary of the invention

[0006] Problems to be solved by the invention

[0007] On the other hand, in recent years, there is a demand to notify a sign of a failure before a device fails. However, in the prior art described in Patent Document 1, although the cause of a device failure can be analyzed after the failure occurs, the sign of a failure cannot be notified.

[0008] Therefore, an object of the present invention is to notify a malfunction sign of a device.

[0009] Means for solving problems

[0010] The device bad sign notification device of the present invention notifies the bad sign of the device, and is characterized in that the device bad sign notification device comprises: an action log acquisition unit, which acquires the action log of the device; a bad sign rule database, which stores bad sign rules that correspond the action log data of the device and the bad sign content of the device; a bad sign determination unit, which determines that there is a bad sign corresponding to the bad sign content when the action log acquired by the action log acquisition unit matches the action log data in the bad sign rule; and a bad sign notification unit, which notifies the bad sign content determined by the bad sign determination unit.

[0011] This makes it possible to detect a malfunction sign of a device by a simple method and notify the content of the malfunction sign.

[0012] In the equipment bad sign notification device of the present invention, it may also be that the equipment bad sign notification device includes an action pattern database, the action pattern database stores an action pattern which is a combination of one or more of the action logs of the equipment and detects the bad sign content of the equipment when the combination is detected, the bad sign rule corresponds the action pattern stored in the action pattern database and the bad sign content of the equipment, the bad sign content includes a predicted bad sign for the equipment and a countermeasure for preventing the predicted bad sign from occurring, and when the bad sign determination unit detects the action pattern included in the bad sign rule, it determines that there is a predicted bad sign included in the bad sign content corresponding to the action pattern and that the countermeasure included in the bad sign content needs to be performed.

[0013] In this way, the bad sign determination unit detects bad sign contents including a bad sign predicted for the device and a response to prevent the predicted bad sign from occurring based on an action pattern formed by combining a plurality of action logs, and the bad sign notification unit notifies the predicted bad sign and the response required to prevent the bad sign from occurring. Therefore, responses can be taken before a bad sign occurs in the device so as not to cause the bad sign predicted for the device, thereby reducing the number of bad sign occurrences.

[0014] In the equipment failure symptom notification device of the present invention, it may also be that the equipment failure symptom notification device includes: a response performance database, which stores the time and response content of the failure response of the equipment as response performance data; and a failure symptom rule generation unit, which generates the failure symptom rule, and the failure symptom rule generation unit sets a first period before a time when a failure response of the equipment is performed and a second period after a time, with reference to the response performance database. When the number of detections of the action mode in the second period is less than a certain ratio of the number of detections of the action mode in the first period, the failure symptom rule generation unit generates the failure symptom rule by matching the action mode, the failure predicted for the equipment and the one response, and stores it in the failure symptom rule database.

[0015] This makes it possible to generate a failure sign rule using a simple method.

[0016] In the device for notifying a bad sign of equipment of the present invention, it may also be that the device for notifying a bad sign of equipment comprises: an action log database storing an action log of the device; an image database storing image data obtained by photographing the device and its surroundings with a camera; an action pattern generating unit generating the action pattern; an image display unit displaying, as an image, image data of a specified time period before and after the moment when the action pattern generated by the action pattern generating unit is detected, from the image data stored in the image database; and a bad sign compliance signal acquiring unit acquiring a bad sign compliance signal to obtain whether the generated action pattern represents a bad sign predicted for the device, and when a bad sign compliance signal is input to the bad sign compliance signal acquiring unit, the action pattern generating unit calculates an evaluation index value of the generated action pattern, and when the calculated evaluation index value is greater than the evaluation index value of the existing action pattern stored in the action pattern database, the generated action pattern is stored as a new action pattern in the action pattern database.

[0017] This makes it possible to generate an operation pattern that can reliably detect a predicted defect.

[0018] The device for notifying a bad sign of an equipment according to the present invention is characterized in that the device comprises: an action log database storing an action log of the equipment; an image database storing image data obtained by photographing the equipment and its surroundings with a camera; a bad sign inference model taking the action log or the action log and the image data as input and taking the bad sign content of the equipment as output; a bad sign inference unit inputting the action log or the action log and the image data into the bad sign inference model to infer the bad sign content; and a bad sign notification unit notifying the bad sign content output by the bad sign inference unit, wherein the bad sign content includes a predicted bad sign for the equipment and a countermeasure for preventing the predicted bad sign from occurring.

[0019] In this way, the bad sign inference unit infers the bad sign content including the bad predicted for the equipment and the response for preventing the predicted bad from occurring based on the action log or the action log and image data, and the bad sign notification unit notifies the predicted bad sign and the response required to prevent the bad from occurring. Therefore, it is possible to take response before the equipment fails so as not to cause the expected bad for the equipment, thereby reducing the number of occurrences of the bad.

[0020] In the device bad sign notification device of the present invention, it can also be that the device bad sign notification device includes: an action pattern database, which stores an action pattern that is a combination of one or more of the action logs of the device and detects the bad sign content of the device when the combination is detected; a response performance database, which stores the time and response content of the bad response of the device as response performance data; and a bad sign inference model generation unit, which generates the bad sign inference model, the bad sign inference model generation unit sets a first period before a moment when the bad response of the device is performed and a second period after a moment, with reference to the response performance database, and the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period. In this case, the bad sign inference model generating unit generates first training data with bad signs which takes the action log or the action log and the image data of the first period as input and outputs the bad sign content, the bad sign inference model generating unit sets a third period which does not overlap with the first period and the second period, and generates second training data without bad signs which takes the action log or the action log and the image data of the third period as input and outputs no bad signs, and the bad sign inference model generating unit generates the bad sign inference model which takes the action log or the action log and the image data of a certain period as input and outputs the bad sign content or no bad signs by learning with the first training data and the second training data.

[0021] As a result, a failure sign inference model can be generated by a simple method, which infers failure sign contents including a failure predicted for the device and a countermeasure for preventing the predicted failure from occurring, based on the action log or the action log and image data.

[0022] In the bad sign notification device of the present invention, it may be that the bad sign inference model generating unit generates the first training data with bad signs by taking the action log and the response performance data or the action log, the response performance data and the image data of the first period as input and the bad sign content as output, and the bad sign inference model generating unit generates the bad sign without bad signs by taking the action log and the response performance data or the action log, the response performance data and the image data of the third period as input and the bad sign content as output. The second training data, the bad sign inference model generating unit generates the bad sign inference model which outputs the bad sign content or no bad sign for the input of the action log and the response performance data or the action log, the response performance data and the image data for a certain period of time by learning with the first training data and the second training data, and the bad sign inference unit inputs the action log and the response performance data or the action log, the response performance data and the image data into the bad sign inference model to infer the bad sign content.

[0023] In this way, the first training data used in learning uses the response performance data as input, so it is possible to associate the past response performance data with the action log for learning. Therefore, for example, it is possible to learn whether the material is inferred to be a bad sign that requires replacement, whether the error is reduced by cleaning, or whether there is no significant change.

[0024] In the equipment bad sign notification device of the present invention, it can also be that the equipment bad sign notification device has: an action pattern generating unit, which generates the action pattern; an image display unit, which displays image data of a specified time period before and after the moment when the action pattern generated by the action pattern generating unit is detected, among the image data stored in the image database, as an image; and a bad sign compliance signal acquisition unit, which obtains a bad sign compliance signal whether the generated action pattern represents a bad sign predicted for the equipment, and when a bad sign compliance signal is input to the bad sign compliance signal acquisition unit, the action pattern generating unit calculates an evaluation index value of the generated action pattern, and when the calculated evaluation index value is greater than the evaluation index value of the existing action pattern stored in the action pattern database, the generated action pattern is stored as a new action pattern in the action pattern database.

[0025] This makes it possible to generate an operation pattern that can reliably detect a predicted defect.

[0026] The action pattern generation method of the present invention is a method for generating an action pattern of a device for detecting a bad sign predicted for the device, characterized in that an action log database storing action logs of the device, an image database storing image data obtained by photographing the device and its surroundings with a camera, and an action pattern database storing action patterns are prepared, an action pattern is generated as a combination of one or more action logs of the device, and a bad sign of the device is detected when the combination is detected, image data of a specified time period before and after the moment when the generated action pattern is detected among the image data stored in the image database is displayed as an image, and it is determined whether the generated action pattern indicates a bad sign predicted for the device based on the displayed image. When it is determined that the image indicates a bad sign, an evaluation index value of the generated action pattern is calculated, and when the calculated evaluation index value is greater than the evaluation index value of the existing action pattern, the generated action pattern is stored as a new action pattern in the action pattern database.

[0027] In this way, by using an image to determine whether or not the operation pattern corresponds to a malfunction sign and then generating the operation pattern, it is possible to generate an operation pattern that can reliably detect a malfunction sign.

[0028] The method for generating bad symptom rules of the present invention is a method for generating bad symptom rules for detecting bad symptom contents of a device in a bad symptom notification device for notifying bad symptom of a device, characterized in that the bad symptom contents include a bad predicted for the device and a response for preventing the predicted bad from occurring, an action pattern database and a response performance database are prepared, the action pattern database stores an action pattern which is a combination of one or more action logs of the device and detects the bad symptom contents of the device when the combination is detected, the response performance database stores the time when the response to the bad of the device was performed and the response content as response performance data, a first period before a time when a response to the bad of the device was performed and a second period after a time are set with reference to the response performance database, and when the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period, the action pattern, the bad predicted for the device and the one response are matched to generate a bad symptom rule.

[0029] This makes it possible to generate a failure sign rule that associates an operation mode with an expected failure of a device and a countermeasure to prevent the predicted failure from occurring.

[0030] The method for generating a bad sign inference model of the present invention is a method for generating a bad sign inference model for inferring the bad sign content of the device in a device bad sign notification device for notifying the bad sign of the device, characterized in that the bad sign content includes a bad predicted for the device and a countermeasure for preventing the predicted bad sign from occurring, and an action log database, an action pattern database, a countermeasure performance database and an image database are prepared, the action log database stores the action log of the device, the action pattern database stores an action pattern which is a combination of one or more action logs of the device and detects the bad sign content of the device when the combination is detected, the countermeasure performance database stores the time and countermeasure content of the bad response of the device as countermeasure performance data, the image database stores image data obtained by photographing the device and the surrounding area of ​​the device using a camera, and a reference is made to the countermeasure performance database to set In a first period before a moment when an adverse response to the device is performed and a second period after a moment, when the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period, first training data with adverse signs is generated, which takes the action log or the action log and the image data of the first period as input and outputs the adverse sign content; a third period is set that does not overlap with both the first period and the second period, and second training data without adverse signs is generated, which takes the action log or the action log and the image data of the third period as input and outputs no adverse signs; and by learning with the first training data and the second training data, an adverse sign inference model is generated, which takes the action log or the action log and the image data of a certain period as input and outputs the adverse sign content or no adverse signs.

[0031] As a result, a failure sign inference model can be generated by a simple method, which infers failure sign contents including a failure predicted for the device and a countermeasure for preventing the predicted failure from occurring, based on the action log or the action log and image data.

[0032] In the method for generating an adverse sign inference model of the present invention, it may be that the first training data with adverse signs is generated by taking the action log and the response performance data or the action log, the response performance data and the image data of the first period as input and the adverse sign content as output, and the second training data without adverse signs is generated by taking the action log and the response performance data or the action log, the response performance data and the image data of the third period as input and the absence of adverse signs as output, and the adverse sign inference model is generated by learning with the first training data and the second training data, with the action log and the response performance data or the action log, the response performance data and the image data of a certain period as input and the adverse sign content or the absence of adverse signs as output.

[0033] In this way, the first training data used in learning uses the response performance data as input, so it is possible to associate the past response performance data with the action log for learning. Therefore, for example, it is possible to learn whether the material is inferred to be a bad sign that requires replacement, whether the error is reduced by cleaning, or whether there is no significant change.

[0034] Effects of the Invention

[0035] The present invention can notify a device of a bad sign. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 1 is a system diagram showing a hardware configuration of the equipment failure sign notification device according to the embodiment.

[0037] Figure 2 This is a functional block diagram of a failure sign notification device according to an embodiment.

[0038] Figure 3 It is shown Figure 2 The diagram shows the data structure of the action log database.

[0039] Figure 4 It is shown Figure 2 A diagram showing the data structure of the action pattern database.

[0040] Figure 5 It is shown Figure 2 The diagram shows the data structure of the failure symptom compliance database.

[0041] Figure 6 It is shown Figure 2 The diagram shows the data structure of the corresponding performance database.

[0042] Figure 7 It is shown Figure 2FIG. 1 is a diagram showing the data structure of the bad symptom rule database.

[0043] Figure 8 : is a flowchart showing the operation of generating an operation pattern.

[0044] Fig. 9 This is an action explanation diagram of the action pattern generation action.

[0045] Fig.10 This is a flowchart showing the operation of generating a failure sign rule.

[0046] Fig.11 This is an operation diagram for explaining the operation of generating a bad sign rule.

[0047] Fig.12 1 is a flowchart showing the failure sign notification operation of the failure sign notification device.

[0048] Fig.13 is a functional block diagram of a fault sign device according to another embodiment.

[0049] Fig.14 This is a flowchart showing the operation of generating a failure sign inference model.

[0050] Fig.15 This is an operation diagram for explaining the generation operation of the adverse sign inference model.

[0051] Fig.16 This is a flowchart showing the failure sign notification operation of the failure sign notification device according to another embodiment.

[0052] Fig.17 This is a flowchart showing another operation of generating a failure sign inference model.

[0053] Fig.18 This is another action explanation diagram of the generation action of the adverse symptom reasoning model.

[0054] Fig.19 This is a flowchart showing another failure sign notification operation of the failure sign notification device according to another embodiment. DETAILED DESCRIPTION

[0055] Next, the device failure sign notification device 100 of the embodiment will be described with reference to the accompanying drawings. Figure 1 As shown, the case where the equipment failure sign notification device 100 is connected to the entrance gate 200 of the room entrance management system 500 will be described. The entrance gate 200 is a target equipment for which the equipment failure sign notification device 100 performs failure sign detection.

[0056] like Figure 1As shown, the room entry management system 500 is composed of the entrance gate 200, the camera 302, and the server 301. The entrance gate 200, the camera 302, and the server 301 are connected via a communication line 303.

[0057] The entrance gate 200 is composed of a reader 201 for reading an ID card 350, a shutter 202 for opening and closing the gate, an external communication interface (hereinafter referred to as I / F) 206 for performing data communication with an external server 301, and a gate control unit 203 for controlling the reader 201, the shutter 202, and the external communication I / F 206. The gate control unit 203 is a computer that internally includes a CPU 204 as a processor for performing information processing and a memory 205 storing programs and control data.

[0058] The server 301 is connected to the external communication I / F 206 of the entrance gate 200 via a communication line 303. The server 301 is composed of a general-purpose computer having a CPU as a processor and a memory (not shown) inside. An authentication database is stored in the memory of the server 301. In addition, an action log of the entrance gate 200 is stored in the memory of the server 301.

[0059] The door control unit 203 outputs the authentication information of the ID card 350 read by the reading unit 201 to the server 301 via the external communication I / F 206 and the communication line 303. The server 301 compares the authentication information of the ID card 350 with the authentication information stored in the authentication database, and when the authentication is successful, sends an authentication success signal to the door control unit 203. After receiving the authentication success signal from the server 301, the door control unit 203 opens the shutter 202.

[0060] The camera 302 captures an image of the entrance gate 200 and its surroundings, and outputs the image to the communication line 303 .

[0061] The equipment failure sign notification device 100 is a computer composed of a CPU 102 as a processor for performing information processing, a storage device 103 for storing control programs, databases, etc., a display 105 for displaying images, a notification device 106 such as an LED and a speaker for notifying alarms, an operator 107 such as a keyboard and a mouse for inputting and outputting data, and an external communication I / F 104 for exchanging data with the outside. The external communication I / F 104 is connected to the entrance gate 200, the server 301, and the camera 302 via a communication line 303 to exchange data.

[0062] The device 100 for notifying a sign of equipment failure comprises: Figure 2 Multiple functional blocks are shown. Figure 1Each functional block is realized by the CPU 102 shown executing a program stored in the storage device 103. In addition, each database is realized by storing data in a predetermined data structure in the storage device 103.

[0063] like Figure 2 As shown, the equipment failure sign notification device 100 includes an image data acquisition unit 11, an image database 12, and an image display unit 13. Here, the image data acquisition unit 11 acquires the image of the entrance gate 200 and the surroundings of the entrance gate 200 obtained by the camera 302. The image database 12 stores the image data acquired by the image data acquisition unit 11. The image database 12 may also continuously store and maintain dynamic images acquired from the camera 302 in a specified period such as the last few days, a week, or a month.

[0064] The image display unit 13 extracts image data of a predetermined period from among the image data stored in the image database 12 , outputs the extracted image data to the display 105 , and causes the display 105 to display a moving image or an image.

[0065] Furthermore, the equipment failure sign notification apparatus 100 includes an operation log acquisition unit 14 , an operation log database 15 , an operation pattern generation unit 16 , an operation pattern database 19 , a failure sign conformity / failure signal acquisition unit 17 , and a failure sign conformity / failure signal database 18 .

[0066] The operation log acquisition unit 14 acquires the operation logs of the reading unit 201, the shutter 202, etc. of the entrance gate 200, and stores them in the operation log database 15. Figure 3 As shown, the action log database 15 is a database that stores date and time and the contents of action logs in association with each other.

[0067] The operation pattern generation unit 16 generates an operation pattern and stores it in the operation pattern database 19. Here, the operation pattern is an operation pattern that is a combination of one or more operation logs and detects the content of a malfunction sign of the entrance gate 200 when the combination is detected.

[0068] Figure 4 Examples of the operation patterns stored in the operation pattern database 19 are shown. Figure 4 The action mode 1 shown will Figure 3 The [Reading Unit] read error at the time 2020 / 10 / 11 05:05:06.920 is set as the action mode. Action mode 2 specifies the following actions: "Action mode 1" occurs within 5 seconds from "Action mode 1", that is, Figure 3As at 2020 / 10 / 11 06:23:05.833 and 06:23:08.090, a [Reader] read error occurs again within 5 seconds after a [Reader] read error occurs. In addition, Action Mode 3 specifies the following event: [Authentication Query] admission permission occurs within 5 seconds from "Action Mode 1", that is, Figure 3 As in the time 2020 / 10 / 11 05:05:06.920~05:05:09.007, within 5 seconds after the occurrence of the [reading unit] reading error, the second ID card 350 is presented, the reading is successful, the authentication is successful, and entry is permitted. Below, action modes 4 and 5 define the interval or number of occurrences of the [reading unit] reading error and the [authentication inquiry] permission to enter. In addition, action mode 6 defines [failed to open the shutter] or [failed to close the shutter], and action mode 7 defines the interval or number of occurrences of [failed to open the shutter] or [failed to close the shutter].

[0069] Thus, the operation modes 1 to 5 are operation modes for detecting a reading failure of the reading unit 201 , and the operation modes 6 and 7 are operation modes for detecting an operation failure of the shutter 202 .

[0070] The action pattern generation unit 16 generates the action pattern described above by combining one or more action logs stored in the action log database 15, and calculates the evaluation index value of the generated action pattern when a bad sign compliance signal is input to the bad sign compliance signal acquisition unit 17. Then, when the calculated evaluation index value is greater than the evaluation index value of the existing action pattern stored in the action pattern database 19, the generated action pattern is stored in the action pattern database 19 as a new action pattern. In this way, the action pattern database 19 is constructed.

[0071] The bad sign coincidence or not signal acquisition unit 17 acquires the bad sign coincidence or not signal from the outside and stores it in the bad sign coincidence or not signal database 18. Here, the bad sign coincidence or not signal is a signal that specifies whether the detection of the action mode is a bad sign based on the dynamic image obtained by the camera 302 when the action mode is detected. The bad sign coincidence or not signal is composed of two signals: a bad sign coincidence signal and a bad sign non-conformance signal.

[0072] like Figure 5 As shown, the failure sign conformity signal database 18 is a database that stores, in association with an action pattern, an image name in a predetermined time period before and after the time when the action pattern is detected, and whether the action pattern conforms to a failure sign or not.

[0073] In addition, the equipment failure sign notification device 100 includes a response performance data acquisition unit 20 and a response performance database 21. Here, "response" refers to maintenance actions required to maintain the functions of the reading unit 201 or the baffle 202, such as cleaning and component replacement. Figure 6 The response performance database 21 is a database that stores the response time and response content in association. The response performance data acquisition unit 20 can store the response performance data input from the operator 107 in the response performance database 21, or can acquire the response performance data from an external data file and store it in the response performance database 21.

[0074] Furthermore, the equipment failure sign notification apparatus 100 includes a failure sign rule generating unit 22 , a failure sign rule database 23 storing the generated failure sign rules, a failure sign determining unit 24 , and a failure sign notification unit 25 .

[0075] The failure sign rule generation unit 22 refers to the response performance database 21 and the operation pattern database 19 , associates the operation pattern with the failure predicted for the entrance gate 200 and the response, generates a failure sign rule, and stores it in the failure sign rule database 23 .

[0076] Figure 7 2 shows an example of the bad sign rules stored in the bad sign rule database 23. As described above, Figure 7 The action mode 4 shown specifies the interval or number of occurrences of a [reading unit] reading error and the occurrence of [authentication inquiry] permission to enter. Bad sign rule 1 associates action mode 4 with bad sign content specified as "bad sign of reading of the reading unit requiring cleaning of the reading unit". Similarly, bad sign rule 2 associates action mode 5 with bad sign content specified as "bad sign of reading of the reading unit requiring replacement of the reading unit". In addition, bad sign rule 3 associates action mode 7 with bad sign content specified as "bad sign of baffle action requiring replacement of the baffle". In this way, the bad sign content includes a predicted badness for the entrance gate 200 and a countermeasure for preventing the predicted badness from occurring.

[0077] When detecting an operation pattern included in a failure sign rule stored in the failure sign rule database 23, the failure sign determination unit 24 determines that a predicted failure sign included in the failure sign content corresponding to the operation pattern exists and that the countermeasure included in the failure sign content is necessary.

[0078] The failure sign notification unit 25 outputs the failure sign content determined by the failure sign determination unit 24 to the notification device 106 for notification.

[0079] Reference Figure 8 to Figure 12The operation of the equipment failure sign notification device 100 configured as described above will be described. Fig.12 As shown, the device bad sign notification device 100 of the embodiment obtains the action log of the entrance gate 200, and when an action pattern matching the action pattern included in the bad sign rule is detected, the bad sign predicted for the entrance gate 200 and the necessary response are notified. Below, before explaining the bad sign detection and notification action, the action of generating the action pattern and the action of generating the bad sign rule are explained.

[0080] First, refer to Figure 8 , Fig. 9 The following describes the operation of generating an action pattern. Figure 8 The initial state before the action shown is stored in the action log database 15. Fig. 9 The action log stores, for example, one month, two to three months, or six months to one year of action logs. In addition, the image database 12 stores dynamic image data for the same period as the period in which the action logs are stored. In addition, the action pattern database 19 stores a plurality of action patterns as initial values. In the following description, it is assumed that there is one action pattern stored. Figure 4 The operation mode 1 shown is used as the initial value to generate the operation modes 2 and 3. Fig. 9 In FIG. 1 , the hollow arrow indicates the detection timing of the action mode 1, the lower right hatched arrow indicates the detection timing of the action mode 2, the black arrow indicates the detection timing of the action mode 3, and the lower left hatched arrow indicates the detection timing of the action mode 4. In addition, Fig. 9 The circled numbers in indicate the action mode numbers.

[0081] like Figure 8 As shown in step S101, the action pattern generation unit 16 generates an action pattern with reference to the action log database 15. The generated action pattern is created as a combination of action logs other than the action pattern 1 stored in the action pattern database 19. The creation method is arbitrary, but, for example, in the case of generating an action pattern for the purpose of detecting poor reading of the reading unit 201, action logs related to poor reading of the reading unit 201 may be extracted from the action logs and randomly combined to generate the action logs. In addition, the action pattern may be generated externally and input from the operator 107.

[0082] In the following description, it is assumed that the action pattern generation unit 16 generates Figure 4 The following description will be made based on the operation modes 2, 3, and 4 shown in the figure. First, the case where the operation mode 2 is generated will be described.

[0083] Next, the action pattern generation unit 16 enters Figure 8In step S102, the action log recorded within a predetermined period is obtained from the action log database 15. The predetermined period is arbitrary, but may be, for example, several hours, several days, or several months.

[0084] Next, the action pattern generation unit 16 enters Figure 8 In step S103, it is determined whether there is a time point where the combination of action logs matches the action pattern 2 in the action logs during the acquired period.

[0085] Action mode 2 specifies that "action mode 1" will occur within 5 seconds from "action mode 1", that is, the action of causing a read error in the reader to occur again within 5 seconds after a read error in the reader occurs. The action mode generation unit 16 repeatedly executes Figure 8 In steps S102 and S103, the action log is scanned in chronological order to find the moment when two consecutive [reading unit] reading errors occur. Fig. 9 In the action log shown, [Reading Unit] read errors occur continuously at time t4 (2019 / 10 / 11 06:23:05.833) and time t5 (2019 / 10 / 11 06:23:08.090). Next, the action pattern generation unit 16 determines whether the interval between time t4 and time t5 is within 5 seconds. The interval between time t4 and time t5 is 2.257 seconds, which is within 5 seconds. Therefore, the action pattern generation unit 16 determines that the combination of the action logs at time t4 and time t5 conforms to action pattern 2. Figure 8 In step S103, it is judged as "yes", and then the Figure 8 At this time, time t5 is the time when the operation mode 2 is detected.

[0086] exist Figure 8 In step S104 , the operation pattern generation unit 16 outputs the time t5 at which the operation pattern 2 is detected to the image display unit 13 .

[0087] exist Figure 8 In step S105 , the image display unit 13 extracts the dynamic image data of a predetermined time period before and after the time t5 from the image database 12 , and displays the dynamic image on the display 105 .

[0088] exist Figure 8 In step S106, the action pattern generation unit 16 determines whether the action pattern 2 indicates a bad sign predicted for the entrance gate 200. This determination can be performed by various means, but as an example, the action pattern generation unit 16 may input a bad sign match signal to the bad sign match or not signal acquisition unit 17, and when the bad sign match signal is input to the action pattern generation unit 16 via the bad sign match or not signal database 18, Figure 8The judgment is "Yes" in step S106.

[0089] Here, the bad sign compliance signal is, for example, a signal for excluding a state in which a reading error of the [reading unit] occurs when an ID card 350 containing inappropriate ID information is presented from the generation of the action mode. For example, in a case where a person presents an ID card 350 but leaves immediately without opening the shutter 202, it can be determined that a reading error of the [reading unit] occurs when an ID card 350 containing inappropriate ID information is presented. On the other hand, in a case in which a reading error of the [reading unit] occurs when the ID card 350 is presented again after a reading error of the [reading unit] occurs, and the ID card 350 is presented again, and the ID authentication succeeds and entry is permitted, it can be determined that there is a bad sign of poor reading of the reading unit 201. This determination can be easily determined by observing a dynamic image captured by the camera 302 disposed near the entrance gate 200.

[0090] Therefore, the image display unit 13 displays the dynamic image 5 (see FIG. 5 ) of the predetermined time period before and after the time t5 at which the operation mode 2 is detected on the display 105. Fig. 9 ), it is determined whether to input the bad sign compliance signal to the bad sign compliance signal acquisition unit 17, or whether to input the bad sign non-compliance signal to the bad sign compliance signal acquisition unit 17. Regarding this determination, an engineer may make the determination when setting the equipment bad sign notification device 100, and input the signal from the operator 107. In addition, for example, AI may make the determination based on the dynamic image data output by the image display unit 13 to the display 105, and input the signal to the bad sign compliance signal acquisition unit 17.

[0091] The action pattern generating unit 16 Figure 8 If the judgment in step S106 is "yes", enter Figure 8 In step S107, the evaluation index value of the generated action pattern 2 is calculated. Here, the evaluation index value is an evaluation index value based on the fit rate. The fit rate is the ratio of the number of times the bad sign matching signal is input when the action pattern is detected to the number of times the action pattern is detected.

[0092] When the fit rate is used as the evaluation index value, Fig. 9 In the example of the operation log database 15 shown, the evaluation index value of the operation pattern 1 is 3 / 4 as described below.

[0093] Operation mode 1 specifies the case where a [reading unit] reading error is detected. Fig. 9In the example shown, the action pattern 1 is detected four times at time t1, time t3, time t4, and time t5. Among them, after the action pattern 1 is detected at time t3, the ID card 350 is not immediately presented again. In this case, the person who left immediately after the ID card 350 was presented is recorded in the dynamic image 3. Therefore, at time t3, even if the action pattern 1 is detected, the bad sign non-conformity signal is input. At other times t1, time t4, and time t5, after the reading error of the [reading unit] as the action pattern 1 is detected, the ID card 350 is presented again and entry is allowed. Therefore, at time t1, time t4, and time t5, the bad conformity signal is input according to the dynamic image 1, dynamic image 4, and dynamic image 5. Therefore, compared with the number of detections of the action pattern 1 four times, the number of times the bad sign conformity signal is input when the action pattern 1 is detected is 3 times, and the ratio is 3 / 4.

[0094] Furthermore, at time t5, action pattern 2 is detected, and a failure sign coincidence signal is input based on the dynamic image 5 at time t5. Therefore, the number of times action pattern 2 is detected is 1, and the number of times the failure sign coincidence signal is input when action pattern 2 is detected is 1, and the ratio is 1 / 1.

[0095] As described above, the evaluation index value of the operation mode 1 is 3 / 4, and the evaluation index value of the operation mode 2 is 1 / 1.

[0096] The action pattern generating unit 16 Figure 8 After calculating the evaluation index value of action mode 2 in step S107, enter Figure 8 In step S108, it is determined whether the evaluation index value of the generated action mode 2 is greater than the evaluation index value of the existing action mode 1. As described above, the evaluation index value of the action mode 2 is 1 / 1, which is greater than 3 / 4 of the evaluation index value of the action mode 1. Therefore, the action mode generating unit 16 generates the action mode 2 in step S108. Figure 8 In step S108, if it is judged as "yes", enter Figure 8 Step S109.

[0097] exist Figure 8 In step S109 , the motion pattern generation unit 16 stores the generated motion pattern 2 as a new motion pattern in the motion pattern database 19 , and ends the motion pattern generation operation.

[0098] In addition, the action pattern generation unit 16 Figure 8 If the judgment is "No" in step S106 or step S108, return to Figure 8 Step S101 generates the next action mode and repeats the process Figure 8 Steps S102 to S109.

[0099] Next, a description will be given of a case where the operation pattern 3 is generated. The description of the same operations as those described above in the operation of generating the operation pattern 2 will be omitted.

[0100] Action mode 3 stipulates that the [authentication inquiry] admission permission occurs within 5 seconds from "action mode 1", that is, the [authentication inquiry] admission permission occurs within 5 seconds after the [reading unit] reading error occurs. As in the case described above, the action mode generation unit 16 repeatedly executes Figure 8 In steps S102 and S103, the action log is scanned in chronological order to find the time when the [Authentication Query] permission to enter occurs within 5 seconds from the reading error of the [Reading Unit]. Fig. 9 In the action log shown, a read error of [Reading Unit] occurs at time t1 (2019 / 10 / 11 05:05:06.920), and then [Authentication Query] admission permission occurs at time t2 (2019 / 10 / 11 05:05:09.007). Moreover, the interval between time t1 and time t2 is 2.087 seconds, which is shorter than 5 seconds. Therefore, the action pattern generation unit 16 determines that the combination of the action logs at time t1 and time t2 conforms to action pattern 3.

[0101] Likewise, in Fig. 9 Between time t5 and time t6, [Authentication Query] admission permission occurs within 5 seconds after the [Reading Unit] reading error occurs. Therefore, the action pattern generation unit 16 determines that the combination of the action logs at time t5 and time t6 conforms to action pattern 3.

[0102] Then, the action pattern generation unit 16 Figure 8 In step S103, it is judged as "yes", and then the Figure 8 In step S104, the time t2 and the time t6 at which the action mode 2 is detected are output to the image display unit 13. The image display unit 13 displays the dynamic image 2 at the time t2 and the dynamic image 6 at the time t6 on the display 105.

[0103] In the case where the [Authentication Query] permission to enter occurs within 5 seconds after the [Reading Unit] reading error occurs as in the operation mode 3, the reading error occurs when the entrant presents the ID card 350, and the ID card 350 is presented again to be permitted to enter. This is recorded in the dynamic image 2 and the dynamic image 6, so at both the time t2 and the time t6, the bad sign coincidence signal is input to the bad sign coincidence signal acquisition unit 17. Then, the operation mode generation unit 16 Figure 8 In step S106, if it is judged as "yes", enter Figure 8 Step S107.

[0104] The action pattern generating unit 16 Figure 8In step S107, the evaluation index value is calculated. In the case of operation mode 3, as described above, the failure sign corresponding signal is input at both the detected times t2 and t6, and therefore the evaluation index value becomes 2 / 2.

[0105] The evaluation index value of the action mode 3 is more than 3 / 4 of the evaluation index value of the existing action mode 1. Figure 8 In step S108, if it is judged as "yes", enter Figure 8 In step S109, the action mode 3 is stored in the action mode database 19 as a new action mode.

[0106] Next, the generation of action mode 4 is described. Figure 4 As shown, action mode 4 is defined as "(authentication inquiry to allow entry) occurs more than 2 times within the last 7 seconds" and (complies with action mode 1), that is, when there is an authentication inquiry to allow entry and more than 2 [reading unit] reading errors have occurred recently. Fig. 9 In the case where a reading error occurs twice at t4 and t5 as shown between time t4 and time t6, and an error occurs at time t6 within 7 seconds of time t4 (authentication inquiry for admission permission), the combination of the action logs at time t4, time t5, and time t6 constitutes action mode 4. Furthermore, action mode 4 is detected at time t6. As described above, at time t6, a bad sign match signal is input to the bad sign match signal acquisition unit 17, and therefore, the action mode generation unit 16 generates an error at time t6. Figure 8 In step S106, the judgment is "yes". In addition, the action mode 4 is similar to the action mode 2. The number of times the bad sign matching signal is input when the action mode 4 is detected is 1, and the ratio is 1 / 1. Therefore, the action mode generation unit 16 Figure 8 In step S108 , it is determined as “Yes”, and the action mode 4 is stored in the action mode database 19 as a new action mode.

[0107] Although the generation of the operation patterns 2, 3, and 4 has been described above, other operation patterns are also generated by the same method and stored in the operation pattern database 19.

[0108] Next, refer to Fig.10 , Fig.11 The generation of the failure sign rule will be described. As described above, the failure sign rule associates the operation pattern stored in the operation pattern database 19 with the failure sign content of the entrance gate 200 as a device. Here, the failure sign content includes the failure predicted for the entrance gate 200 and the countermeasures for preventing the predicted failure from occurring.

[0109] like Fig.11As shown, the response time and the response content of the entrance gate 200 for about half a year to one year are stored in association in the response performance database 21 .

[0110] like Fig.10 As shown in step S201, the bad sign rule generation unit 22 determines whether there is a record of cleaning, component replacement, etc. of the entrance gate 200 in the record database 21. Then, the bad sign rule generation unit 22 Fig.10 If the judgment in step S201 is "yes", enter Fig.10 Step S202, such as Fig.11 As shown in FIG. 1 , a time t11 at which an adverse response is performed is extracted, and a 1A period before the time t11 and a 2A period after the time t11 are set. Fig.10 When it is determined as “NO” in step S201 , the failure-sign rule generating unit 22 ends the operation of generating the failure-sign rule.

[0111] Then, in Fig.10 In step S203, the bad sign rule generating unit 22 obtains the action log of the 1A period from the action log database 15. Fig.10 In step S204, it is determined whether the combination of the acquired action logs matches any one of the action patterns stored in the action pattern database 19. Fig.11 As shown, in the 1A period, the operation mode 1 and the operation mode 4 are detected. When the operation mode is detected, the failure sign rule generating unit 22 generates the operation mode 1 and the operation mode 4 in the 1A period. Fig.10 In step S204, if it is judged as "yes", enter Fig.10 Step S205. Fig.10 If the determination result in step S204 is "NO", the failure-sign rule generating unit 22 ends the operation of generating the failure-sign rule.

[0112] exist Fig.10 In step S205, the failure sign rule generating unit 22 determines whether the ratio of the number of detections of the operation pattern 1 in the 2A period to the number of detections of the operation pattern 1 in the 1A period is less than a certain ratio. Here, the certain ratio can be freely set, but can be set to less than 10%, for example.

[0113] In the 1A period, the action pattern 1 is detected twice, and in the 2A period, the action pattern 1 is detected once. Then, the above ratio becomes 50%, which is larger than the set value of 10%. Therefore, the failure sign rule generation unit 22 Fig.10 In step S205, if it is judged as "No", the Fig.10 In step S207, it is determined whether there is an operation mode other than the operation mode 1 during the 1A period.

[0114] like Fig.11 As shown in FIG. 1A, the operation mode 4 is detected during the 1A period, so the bad sign rule generating unit 22 generates the bad sign rule 4 in the Fig.10 In step S207, it is judged as "yes", and the Fig.10 In step S205, it is determined whether the above ratio of the operation mode 4 is less than a certain ratio.

[0115] like Fig.11 As shown in FIG. 1A, the action pattern 4 is detected twice during the 1A period, and the action pattern 4 is not detected during the 2A period. Therefore, the above ratio becomes 0%, which becomes a certain ratio of 10% or less. Therefore, the bad sign rule generating unit 22 Fig.10 If the judgment in S205 is "yes", the Fig.10 Step S206.

[0116] exist Fig.10 In step S206, the bad sign rule generating unit 22 associates the action mode 4 with the predicted bad condition and the response for the entrance gate 200, and generates a bad sign rule. The response at time t11 is to clean the reader, and the action mode 4 involves bad reading of the reader. Therefore, Figure 7 As shown, the fault sign rule generating unit 22 sets the fault sign content corresponding to the operation mode 4, such as "the fault sign of the reading unit needing cleaning of the reading unit", and generates the fault sign rule 1. Then, the fault sign rule generating unit 22 stores the generated fault sign rule 1 in the fault sign rule database 23.

[0117] Then, the bad sign rule generating unit 22 enters Fig.10 In step S207, it is determined whether there is an action mode other than action modes 1 and 4 in period 1A. Fig.11 As shown in FIG. 1A, no operation mode other than operation mode 1 and operation mode 4 is detected during the 1A period. Therefore, the failure sign rule generating unit 22 generates the failure sign rule in FIG. Fig.10 In step S207, if it is judged as "No", the Fig.10 In step S208, it is determined whether there are other moments when bad processing is performed.

[0118] like Fig.11 As shown in FIG. 1 , except for the time t11, the entrance gate 200 is handled four times at t12 to t15. Therefore, the fault sign rule generating unit 22 generates the fault sign rule at Fig.10 In step S208, it is judged as "yes", and the Fig.10 In step S202, the same action is performed at time t12.

[0119] In the 1B period before time t11, action mode 1, action mode 6, and action mode 7 are detected. Action mode 1 is detected once in the 1B period and once in the 2B period. Therefore, the ratio of the number of detections of action mode 1 in the 2B period to the number of detections of action mode 1 in the 1B period is 100%, which is not a constant ratio of 10% or less. Therefore, the bad sign rule generation unit 22 detects action mode 1 in the 2B period. Fig.10 In step S205, the judgment is "No", Fig.10 In step S207, it is judged as "yes", and the Fig.10 In step S205, the above ratio determination is performed on the action mode 6. The action mode 6 is detected twice in the 1B period and once in the 2B period, so the above ratio becomes 50%, which is not a constant ratio of 10% or less. Fig.10 In step S205, the judgment is "No", Fig.10 In step S207, it is judged as "yes", and the Fig.10 In step S205, the above-mentioned ratio judgment is performed on action mode 7.

[0120] Since the action pattern 7 is detected twice in the 1B period and the action pattern 7 is not detected in the 2B period, the above ratio becomes 0%, which is less than the certain ratio of 10%. Therefore, the bad sign rule generating unit 22 enters Fig.10 In step S206, the action mode 7, the predicted failure and the response for the entrance door 200 are matched to generate a failure sign rule. The response at time t12 is to replace the baffle. The action mode 7 involves the failure of the baffle 202. Therefore, Figure 7 As shown, the failure sign rule generation unit 22 sets the failure sign content corresponding to the operation mode 7, such as "fender operation failure sign requiring replacement of the fender", and generates the failure sign rule 2. Then, the failure sign rule generation unit 22 stores the generated failure sign rule 2 in the failure sign rule database 23.

[0121] The bad sign rule generating unit 22 Fig.10 In step S208, it is judged as "yes", and the Fig.10 In step S202, the 1C period and the 2C period are set around time t13, and the same action as above is performed. In the 1C period, the action mode 1 and the action mode 4 are detected, but the above ratio of any action mode is not less than a certain ratio. Therefore, in Fig.10 The determination in step S205 is "No", and the failure sign rule is not set. The same is true in the 1D period and the 2D period at the next time t14.

[0122] In the 1E period before time t15, action patterns 1, 4, 5, and 6 are detected. Among them, action pattern 5 is detected twice in the 1E period, and action pattern 5 is not detected in the 2E period, so the above ratio becomes 0%, which is a certain ratio of 10% or less. Therefore, as shown in FIG. 23, the bad sign rule generation unit 22 generates a bad sign rule 3 that associates action pattern 5 with the bad sign content of "reading unit reading bad sign that needs to replace the reading unit", and stores it in the bad sign rule database 23.

[0123] As described above, when the number of times the motion pattern is detected after the response becomes less than a certain ratio relative to the number of times the motion pattern is detected in the first period before the response, an adverse sign rule is generated that corresponds the motion pattern and the response content. Therefore, when the motion pattern is detected, the predicted adverse event for the entrance gate 200 and the response for preventing the predicted adverse event from occurring can be output.

[0124] Next, refer to Fig.12 The following describes an operation of notifying a predicted failure of the entrance gate 200 using the failure sign rule and taking measures to prevent the failure. This operation may be performed once every 2 to 3 months or monthly.

[0125] like Fig.12 As shown in step S301, the failure sign determination unit 24 obtains the operation logs of a predetermined period such as 2 to 3 months or 1 month from the operation log database 15. Fig.12 In step S302, it is determined whether the combination of the obtained action logs matches the action pattern included in the bad sign rule. Fig. 9 At time t6 shown, action mode 4 is detected. Figure 7 As shown, action mode 4 is an action mode included in bad sign rule 1. Therefore, the bad sign determination unit 24 determines that the combination of action logs obtained matches the action mode included in bad sign rule 1, and determines "yes" in step S302, and enters Fig.12 Then, in step S303. Fig.12 In step S303, the bad sign determination unit 24 determines that there is a predicted bad sign included in the bad sign content corresponding to the action mode 4 and that the countermeasures included in the bad sign content are required. Figure 7 As shown, the failure sign determination unit 24 determines that there is a reading failure sign of the reading unit 201 that requires cleaning of the reading unit 201 of the entrance door 200. Then, the failure sign determination unit 24 outputs the determination result to the failure sign notification unit 25.

[0126] exist Fig.12In step S304, the failure sign notification unit 25 notifies the user of the determination result input from the failure sign determination unit 24 using the notification device 106. Fig.12 If the determination result in step S302 is "No", the notification action of the bad sign is terminated.

[0127] As described above, the failure sign determination unit 24 detects failure sign contents including a failure predicted for the entrance gate 200 and a response to prevent the predicted failure from occurring based on an action pattern formed by combining a plurality of action logs, and the failure sign notification unit 25 notifies the predicted failure sign and the response required to prevent the failure from occurring. Therefore, it is possible to take response measures that do not cause the predicted failure for the equipment before the failure occurs in the entrance gate 200, thereby reducing the number of failure occurrences.

[0128] In the above description, an action pattern is generated based on an action log, and a bad sign rule is described by corresponding the action pattern and the bad sign content. However, an action pattern may not be generated, and one action log or a combination of multiple action logs may be set as action log data, and the action log data and the bad sign content may be corresponded as the bad sign rule.

[0129] For example, the action log data such as "Reading unit" reads wrong twice can be matched with the bad sign content such as "Reading unit reads bad sign" and set as the bad sign rule. In this case, it is difficult to detect the corresponding content as the bad sign content, but the bad sign can be notified with a simple structure.

[0130] In the above description, it is assumed that in the action of generating the action pattern, a dynamic image of the period before and after the moment when the action pattern obtained by the camera 302 is detected is displayed on the display 105, and the engineer determines whether to use the dynamic image in the action pattern. Figure 8 Whether the action pattern generated in step S106 indicates a predicted sign of a defect for the entrance door 200 is explained. However, at this time, the engineer can know from the dynamic image that a defect has occurred in the entrance door 200, and can also take measures to eliminate the defect if it is determined that a response is necessary.

[0131] Next, refer to Figure 13 to Figure 16 Another embodiment of the device bad sign notification device 150 is described. The device bad sign notification device 150 uses a bad sign inference model that takes action logs or action logs and image data as input and outputs the bad sign content of the device to infer the bad sign content of the device, and notifies the inferred bad sign content. Figures 1 to 12 The same parts of the described equipment failure sign notification device 100 are denoted by the same reference numerals, and the description thereof will be omitted.

[0132] like Fig.13 As shown, the device failure sign notification device 150 replaces the reference Figure 1 The equipment failure sign notification apparatus 100 described above includes the failure sign rule generation unit 22 , the failure sign rule database 23 , and the failure sign determination unit 24 , and includes a failure sign inference model generation unit 26 , a failure sign inference model database 27 , and a failure sign inference unit 28 .

[0133] The bad sign inference model generation unit 26 refers to the action log database 15 and the image database 12, and generates a bad sign inference model that outputs bad sign content or no bad sign for the action log or action log and image data input for a certain period of time through learning. The bad sign inference model database 27 is a database that stores the bad sign inference model generated by the bad sign inference model generation unit 26. The bad sign inference unit 28 uses the bad sign inference model to infer the bad sign content, and outputs the inference result to the bad sign notification unit 25.

[0134] Before describing the failure sign notification operation of the equipment failure sign notification device 150, refer to Fig.14 , Fig.15 The generation of the bad sign reasoning model is described below. As described above in the generation of the bad sign rule, Fig.11 As shown, the response time and the response content of the entrance gate 200 for about half a year to one year are stored in association in the response performance database 21 .

[0135] like Fig.14 As shown in step S401, the fault sign inference model generation unit 26 determines whether there is a record of cleaning, component replacement, etc. of the entrance door 200 in the record database 21. Then, the fault sign inference model generation unit 26 Fig.14 If the judgment in step S401 is "yes", enter Fig.14 Step S402, extracting a time t21 at which an adverse response is performed (refer to Fig.15 ), set the first period before time t21 and the second period after time t21. Fig.14 When it is determined as “NO” in step S401, the failure-sign-inference-model generating unit 26 ends the operation of generating the failure-sign-inference-model.

[0136] Then, in Fig.14 In step S403, the failure sign inference model generation unit 26 obtains the action log of the first period from the action log database 15. Fig.14In step S404, it is determined whether the combination of the acquired action logs matches any one of the action patterns stored in the action pattern database 19. Fig.15 As shown, in the first period, the operation mode 1, the operation mode 4, the operation mode 5 and the operation mode 6 are detected. When the operation mode is detected, the failure sign inference model generation unit 26 generates the failure sign inference model. Fig.14 In step S404, if it is judged as "yes", enter Fig.14 Step S405. In addition, Fig.14 When it is determined as “NO” in step S404, the failure-sign-inference-model generating unit 26 ends the operation of generating the failure-sign-inference-model.

[0137] exist Fig.14 In step S405, the failure sign inference model generator 26 determines whether the ratio of the number of detections of the action pattern 1 in the second period to the number of detections of the action pattern 1 in the first period is less than a certain ratio. Here, the certain ratio can be freely set, but can be set to less than 10%, for example.

[0138] In the first period, the action pattern 1 is detected twice, and in the second period, the action pattern 1 is detected twice. Then, the above ratio becomes 100%, which is greater than the set value of 10%. Therefore, the failure sign inference model generation unit 26 Fig.14 In step S405, the judgment is "No", enter Fig.14 In step S410, it is determined whether there is an action mode other than action mode 1 in the first period. In the first period, action modes 4, 5, and 6 are detected in addition to action mode 1. Therefore, the failure sign inference model generation unit 26 generates the failure sign inference model in the first period. Fig.14 In step S410, if it is judged as "yes", return Fig.14 In step S405, regarding operation mode 4, similarly to operation mode 1, it is determined whether the above-mentioned ratio is less than a certain ratio.

[0139] like Fig.15 As shown in FIG. 1 , in the first period, the action pattern 4 is detected three times, and in the second period, the action pattern 4 is detected once. Therefore, in the action pattern 4, the above ratio becomes 33%, which is not the constant ratio of 10% or less. Therefore, the failure sign inference model generation unit 26 Fig.14 In step S405, the judgment is "No", Fig.14 In step S410, if it is judged as "yes", return Fig.14 In step S405, regarding operation mode 5, similarly to operation modes 1 and 4, it is determined whether the above-mentioned ratio is less than a certain ratio.

[0140] like Fig.15As shown in FIG. 1 , in the first period, the action pattern 5 is detected three times, but in the second period, the action pattern 5 is not detected. Therefore, the above ratio becomes 0%, which becomes less than the certain ratio of 10%. Therefore, the failure sign inference model generation unit 26 Fig.14 In step S405, if it is judged as "yes", enter Fig.14 Step S406.

[0141] exist Fig.14 In step S406, Fig.15 As shown, the bad sign inference model generation unit 26 generates the first training data with bad signs, which takes the action log or the action log and image data of the first period as input and outputs the bad sign content. Figure 4 As shown, action mode 5 is defined as "(complying with action mode 4) occurs and (complying with action mode 4) is more than 5 times within the last 60 days". That is, action mode 5 is defined as action mode 4 defined by "(authentication inquiry permission to enter) occurs and [reading unit] reading error is more than 2 times within the last 7 seconds" occurs more than 5 times within the last 60 days. In this way, action mode 5 represents a sign of occurrence of poor reading in the reading unit. In addition, the response content at time t21 is the replacement process of the reading unit. Therefore, the bad sign inference model generation unit 26 defines the bad sign content of the first training model as "poor reading sign of the reading unit that requires replacement of the reading unit". Then, the bad sign inference model generation unit 26 enters Fig.14 In step S407, the third period is set. Fig.15 As shown, the third period is a period that does not overlap with either the first period or the second period.

[0142] Then, the bad sign inference model generation unit 26 enters Fig.14 Step S408, such as Fig.15 As shown, the second training data without bad signs is generated, which takes the action log or the action log and image data of the third period as input and outputs no bad signs. Then, after the generation of the second training data is completed, enter Fig.14 Step S409.

[0143] exist Fig.14 In step S409, the bad sign inference model generation unit 26 generates a bad sign inference model that outputs bad sign content or no bad sign for the input of action logs or action logs and image data for a certain period of time by learning using the first training data and the second training data. Then, the bad sign inference model generation unit 26 stores the generated bad sign inference model in the bad sign inference model database 27.

[0144] The bad sign inference model generation unit 26 enters Fig.14In step S410, it is determined whether there are other action modes. Since action mode 6 still remains, the bad sign inference model generation unit 26 generates the bad sign inference model. Fig.14 In step S410, if it is judged as "No", the system returns Fig.14 In step S405, it is determined that the above ratio of the operation mode 6 is less than a certain ratio. Fig.15 As shown in FIG. 1 , the action pattern 6 is detected once in the first period and once in the second period. Therefore, the above ratio becomes 100%, not a constant ratio of 10% or less. Therefore, the failure sign inference model generation unit 26 Fig.14 In step S405, the judgment is "No". In addition, the action mode 6 is the last action mode for performing the above-mentioned ratio judgment. Fig.14 In step S410, if it is judged as "No", the Fig.14 Then, the bad sign inference model generation unit 26 determines whether there is a time when other measures are taken. If there is a time when other measures are taken, it returns to step S411. Fig.14 Step S402, repeatedly perform the same actions as above. In addition, in the case where there is no other time when other responses are performed, Fig.14 If the judgment in step S411 is "No", the generation action of the failure sign inference model is terminated.

[0145] Next, refer to Fig.16 The operation of notifying the bad sign content using the bad sign inference model will be described. Similar to the case of notifying the bad sign content using the bad sign rule described above, this operation can be performed once every 2 to 3 months or every month.

[0146] like Fig.16 As shown in step S501, the failure sign inference unit 28 obtains action logs of a specified period such as 2 to 3 months or 1 month from the action log database 15, or obtains action logs of a specified period from the action log database 15, and obtains image data of a specified period from the image database 12.

[0147] like Fig.16 As shown in step S502, the bad sign inference unit 28 reads the bad sign inference model from the bad sign inference model database 27, inputs the action log of a specified period or the action log and image data of a specified period into the bad sign inference model, and causes it to infer the bad sign content. Then, when the bad sign inference unit 28 outputs the inference result of bad content, Fig.16 In step S503, if it is judged as "yes", enter Fig.16 Step S504.

[0148] exist Fig.16 In step S504 , the failure sign notification unit 25 notifies the failure sign output from the failure sign inference unit 28 through the notification device 106 .

[0149] When the bad sign inference unit 28 outputs the inference result that there is no bad sign without outputting bad content, Fig.16 In step S503, it is judged as "No", and the notification action of the bad sign is terminated.

[0150] Here, the content of the failure sign notified includes the failure predicted for the entrance gate 200 and the countermeasures for preventing the predicted failure, as specified in the case of generating the first training model, such as the "failure sign of the reading unit of the reading unit requiring replacement of the reading unit". Therefore, it is possible to take countermeasures to prevent the predicted failure from occurring before the failure of the entrance gate 200 occurs, thereby reducing the number of failures.

[0151] Next, refer to Figure 17 to Figure 19 The following describes the operation of generating another failure sign inference model and the inference and notification of failure sign content using another failure sign inference model. Figure 14 to Figure 16 The same operations as the described generation operation of the failure sign inference model and the inference and notification of the failure sign content using the failure sign inference model are denoted by the same reference numerals, and the description thereof will be omitted.

[0152] like Fig.17 As shown in the figure, another method for generating an adverse symptom inference model is Fig.14 In step S406, step S408, and step S409, the corresponding performance data is added to the input items.

[0153] exist Fig.17 In step S601, Fig.18 As shown, the bad sign inference model generation unit 26 generates the first training data with bad signs, which takes the action log and the actual response data or the action log, the actual response data and the image data of the first period as input and outputs the bad sign content. Fig.17 In step S602, Fig.18 As shown, the bad sign inference model generation unit 26 generates the second training data without bad signs, which takes the action log and the actual response data or the action log, the actual response data and the image data of the third period as input and outputs no bad signs. Fig.17 In step S603, Fig.18As shown, the bad sign inference model generation unit 26 generates another bad sign inference model that outputs bad sign content or no bad sign for the input of action logs and response performance data or action logs, response performance data, and image data for a certain period of time by learning using the first training data and the second training data. The bad sign inference model generation unit 26 stores the generated other bad sign inference model in the bad sign inference model database 27.

[0154] In this way, the first training data used in learning uses the response performance data as input, so it is possible to associate the past response performance data with the action log for learning. Therefore, for example, it is possible to learn whether the material is inferred to be a bad sign that requires replacement, whether the error is reduced by cleaning, or whether there is no significant change. Thus, another bad sign inference model is generated as a model that can separately infer whether it is a bad sign that requires replacement or a bad sign that requires cleaning.

[0155] Reference Fig.19 The operation of inferring and notifying the content of a bad sign using another bad sign inference model generated as described above will be described. Fig.16 The same operations as those described above are denoted by the same reference numerals, and the description thereof will be omitted.

[0156] like Fig.19 As shown in step S701, the failure sign inference unit 28 obtains the action logs of the prescribed period from the action log database 15, and obtains the actual response data of the prescribed period from the actual response database 21. Alternatively, the failure sign inference unit 28 obtains the action logs of the prescribed period from the action log database 15, obtains the actual response data of the prescribed period from the actual response database 21, and obtains the image data of the prescribed period from the image database 12.

[0157] like Fig.19 As shown in step S702, the bad sign inference unit 28 reads another bad sign inference model from the bad sign inference model database 27, inputs the action log and response performance data of a specified period or the action log, response performance data and image data of a specified period into the other bad sign inference model, and makes it infer the bad sign content. Then, when the bad sign inference unit 28 outputs the inference result of bad content, Fig.19 In step S503, it is judged as "yes", Fig.19 In step S504, the bad sign content is notified.

[0158] The above-described operation of inferring and notifying the content of a failure sign using another failure sign inference model exerts the same effect as the above-described operation of inferring and notifying based on the failure sign inference model.

[0159] In the above description, the device bad sign notification device 100 that uses bad sign rules to detect bad content and notifies the detected bad sign content, and the device bad sign notification device 150 that uses a bad sign inference model to infer bad sign content and notifies the inferred bad sign content are described. However, it is also possible to set up a device bad sign notification device that can perform both the action of detecting bad content using bad sign rules and notifying the detected bad sign content, and the action of inferring bad sign content using a bad sign inference model and notifying the inferred bad sign content.

[0160] Description of symbols

[0161] 11: Image data acquisition unit; 12: Image database; 13: Image display unit; 14: Action log acquisition unit; 15: Action log database; 16: Action mode generation unit; 17: Bad symptom compliance signal acquisition unit; 18: Bad symptom compliance signal database; 19: Action mode database; 20: Response performance data acquisition unit; 21: Response performance database; 22: Bad symptom rule generation unit; 23: Bad symptom rule database; 24: Bad symptom determination unit; 25: Bad symptom notification unit; 26: Bad symptom inference model generation unit Department; 27: bad sign reasoning model database; 28: bad sign reasoning department; 100, 150: equipment bad sign notification device; 102, 204: CPU; 103: storage device; 104, 206: external communication I / F; 105: display; 106: notifier; 107: operator; 200: entrance door; 201: reading unit; 202: baffle; 203: door control unit; 205: memory; 301: server; 302: camera; 303: communication line; 350: ID card; 500: room entry management system.

Claims

1. A device for notifying a bad sign of equipment, the device notifying a bad sign of equipment, characterized in that: The device for notifying the equipment of a bad sign has the following features: an action log acquisition unit configured to acquire an action log of the device; a bad sign rule database storing bad sign rules for matching the action log data of the device with the bad sign content of the device; a bad sign determination unit configured to determine that there is a bad sign corresponding to the bad sign content when the action log acquired by the action log acquisition unit matches the action log data in the bad sign rule; a bad sign notification unit configured to notify the bad sign content determined by the bad sign determination unit; an action pattern database storing an action pattern which is a combination of one or more of the action logs of the device and detects the bad sign content of the device when the combination is detected; a response performance database storing, as response performance data, the time when the response to the failure of the device was performed and the content of the response; as well as a bad sign rule generating unit, which generates the bad sign rule, The bad sign rule associates the action mode stored in the action mode database with the bad sign content of the device. The fault sign content includes the predicted fault of the device and the countermeasures for preventing the predicted fault from occurring. When the bad sign determination unit detects the action pattern included in the bad sign rule, it determines that there is a predicted bad sign included in the bad sign content corresponding to the action pattern and that a countermeasure included in the bad sign content is required. The failure sign rule generating unit sets a first period before a time when a response to a failure of the equipment is performed and a second period after a time by referring to the response performance database. When the number of detections of the action mode in the second period is less than a certain ratio of the number of detections of the action mode in the first period, the bad sign rule generation unit generates the bad sign rule by matching the action mode, the predicted badness for the device and the one response, and stores it in the bad sign rule database.

2. The device for notifying a sign of equipment failure according to claim 1, characterized in that: The device for notifying of a sign of equipment failure comprises: an action log database storing the action logs of the device; An image database storing image data obtained by photographing the device and its surroundings using a camera; an action pattern generating unit that generates the action pattern; an image display unit that displays, among the image data stored in the image database, image data of a predetermined time period before and after a time when the action pattern generated by the action pattern generating unit is detected, as an image; as well as a failure sign conformity / nonconformity signal acquisition unit for acquiring a failure sign conformity / nonconformity signal indicating whether the generated operation mode indicates a failure sign predicted for the device, When the bad sign compliance signal is input to the bad sign compliance signal acquisition unit, the action mode generation unit calculates the evaluation index value of the generated action mode, and when the calculated evaluation index value is greater than the evaluation index value of the existing action mode stored in the action mode database, the generated action mode is stored as a new action mode in the action mode database.

3. A device for notifying a bad sign of equipment, the device notifying a bad sign of equipment, characterized in that: The device for notifying the equipment of a bad sign has the following features: an action log database storing the action logs of the device; An image database storing image data obtained by photographing the device and its surroundings using a camera; a bad sign inference model, which takes the action log or the action log and the image data as input and takes the bad sign content of the device as output; a bad sign inference unit that inputs the action log or the action log and the image data into the bad sign inference model to infer the bad sign content; a bad sign notification unit configured to notify the bad sign content output by the bad sign inference unit; an action pattern database storing an action pattern which is a combination of one or more of the action logs of the device and detects the bad sign content of the device when the combination is detected; a response performance database storing, as response performance data, the time when the response to the failure of the device was performed and the content of the response; as well as a bad sign inference model generating unit, which generates the bad sign inference model, The fault sign content includes the predicted fault of the device and the countermeasures for preventing the predicted fault from occurring. The failure sign inference model generation unit sets a first period before a time when the failure response of the equipment is performed and a second period after a time by referring to the response performance database. When the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period, the bad sign inference model generation unit generates the first training data with bad signs, which takes the action log of the first period or the action log and the image data as input and takes the bad sign content as output. The bad sign inference model generation unit sets a third period that does not overlap with the first period and the second period, generates second training data without bad signs that takes the action log or the action log and the image data of the third period as input and outputs no bad signs, The failure sign inference model generation unit generates the failure sign inference model that outputs the failure sign content or no failure sign for the input of the action log or the action log and the image data for a certain period of time by learning using the first training data and the second training data.

4. The device for notifying a sign of equipment failure according to claim 3, characterized in that: The failure sign inference model generation unit generates the first training data with failure signs, which takes the action log and the actual response data or the action log, the actual response data and the image data during the first period as input and outputs the failure sign content. The failure sign inference model generation unit generates the second training data without failure signs, which takes the action log and the actual response data or the action log, the actual response data and the image data of the third period as input and outputs no failure signs. The bad sign inference model generation unit generates the bad sign inference model that outputs the bad sign content or no bad sign for the input of the action log and the response performance data or the action log, the response performance data and the image data for a certain period of time by learning with the first training data and the second training data, The failure sign inference unit inputs the operation log and the actual handling data or the operation log, the actual handling data, and the image data into the failure sign inference model, and infers the failure sign content.

5. The device for notifying a sign of equipment failure according to claim 3 or 4, characterized in that: The device for notifying of a sign of equipment failure comprises: an action pattern generating unit that generates the action pattern; an image display unit that displays, among the image data stored in the image database, image data of a predetermined time period before and after a time when the action pattern generated by the action pattern generating unit is detected, as an image; as well as a failure sign conformity / nonconformity signal acquisition unit for acquiring a failure sign conformity / nonconformity signal indicating whether the generated operation mode indicates a failure sign predicted for the device, When the bad sign compliance signal is input to the bad sign compliance signal acquisition unit, the action mode generation unit calculates the evaluation index value of the generated action mode, and when the calculated evaluation index value is greater than the evaluation index value of the existing action mode stored in the action mode database, the generated action mode is stored as a new action mode in the action mode database.

6. A method for generating an operation pattern of a device for detecting a predicted bad sign for the device, characterized in that: preparing an action log database storing an action log of the device, an image database storing image data obtained by photographing the device and its surroundings using a camera, and an action pattern database storing action patterns, generating the operation pattern as a combination of one or more operation logs of the device, and detecting a malfunction sign of the device when the combination is detected, displaying, as an image, image data of a predetermined time period before and after a time when the generated action pattern is detected, from among the image data stored in the image database; determining, based on the displayed image, whether the generated action pattern indicates a predicted bad sign for the device, When it is determined that the image represents a bad sign, an evaluation index value of the generated action mode is calculated, and when the calculated evaluation index value is greater than the evaluation index value of the existing action mode, the generated action mode is stored as a new action mode in the action mode database, and the evaluation index value is the ratio of the number of bad sign matching signals input when the action mode is detected to the number of detections of the action mode.

7. A method for generating a bad sign rule, the bad sign rule being used to detect bad sign content of a device in a bad sign notification device for notifying a bad sign of the device, characterized in that: The fault sign content includes the predicted fault of the device and the countermeasures for preventing the predicted fault from occurring. An action pattern database and a response performance database are prepared, wherein the action pattern database stores an action pattern for detecting the bad sign content of the device when the combination is detected, which is a combination of one or more action logs of the device, and the response performance database stores the time and response content of the bad response of the device as response performance data, referring to the response performance database, setting a first period before a time when a response to a failure of the equipment is performed and a second period after a time; Obtaining the action logs of the first period from the action log database, and when the combination of the obtained action logs matches any one of the action patterns stored in the action pattern database, determining whether the ratio of the number of detections of the action pattern in the second period to the number of detections of the action pattern in the first period is less than a certain ratio, When the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period, a failure sign rule is generated by associating the action pattern, the failure predicted for the device, and the one countermeasure.

8. A method for generating a bad sign inference model, the bad sign inference model being used to infer bad sign content of the device in a bad sign notification device for notifying a bad sign of the device, characterized in that: The fault sign content includes the predicted fault of the device and the countermeasures for preventing the predicted fault from occurring. An action log database, an action pattern database, a response performance database, and an image database are prepared, wherein the action log database stores the action logs of the device, the action pattern database stores an action pattern that is a combination of one or more action logs of the device and detects the bad sign content of the device when the combination is detected, the response performance database stores the time and response content of the bad response of the device as response performance data, and the image database stores image data obtained by shooting the device and the surrounding area of ​​the device with a camera, Referring to the countermeasure performance database, a first period before a time when countermeasures for failure of the equipment are performed and a second period after a time are set; When the number of detections of the action pattern in the second period is less than a certain ratio of the number of detections of the action pattern in the first period, generating first training data with bad signs which takes the action log of the first period or the action log and the image data as input and takes the bad sign content as output, A third period is set that does not overlap with the first period and the second period, and second training data with no bad signs is generated, which takes the action log or the action log and the image data of the third period as input and outputs no bad signs, By learning using the first training data and the second training data, a failure sign inference model is generated which outputs the failure sign content or no failure sign for the input of the action log or the action log and the image data for a certain period of time.

9. The method for generating an adverse sign inference model according to claim 8, characterized in that: generating the first training data with a bad sign, which takes the action log and the actual response data or the action log, the actual response data and the image data during the first period as input and takes the bad sign content as output, generating the second training data without a bad sign, which takes the action log and the actual response data or the action log, the actual response data and the image data of the third period as input and outputs no bad sign, By learning using the first training data and the second training data, the bad sign inference model is generated for the input of the action log and the response performance data or the action log, the response performance data and the image data for a certain period of time, and the bad sign content or the absence of bad signs as output.